Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
ViewVC logotype

Diff of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 937, Thu Sep 20 21:20:28 2001 UTC revision 1037, Sat Jan 26 02:27:14 2002 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16  Name: Matthias Blume  Name: Matthias Blume
17    Date: 2001/01/25 21:27:00 EST
18    Tag: Release_110_38
19    Description:
20    
21    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
22    
23    README will be added later.
24    
25    ----------------------------------------------------------------------
26    Name: Matthias Blume
27    Date: 2002/01/25
28    Tag: blume-20020125-ffi
29    Description:
30    
31    A large number of tweaks and improvements to ml-nlffi-lib and
32    ml-nlffigen:
33    
34       - ML represenation types have been streamlined
35       - getter and setter functions work with concrete values, not abstract
36         ones where possible
37       - ml-nlffigen command line more flexible (see README file there)
38       - some bugs have been fixed (hopefully)
39    
40    ----------------------------------------------------------------------
41    Name: Lal George
42    Date: 2002/01/24
43    Tag: george-20020124-risc-ra-interface
44    Description:
45    
46       There is a dramatic simplification in the interface to the
47       register allocator for RISC architectures as a result of making
48       parallel copy instructions explicit.
49    
50    ----------------------------------------------------------------------
51    Name: Matthias Blume
52    Date: 2002/01/22
53    Tag: blume-20020122-x86-ccalls
54    Description:
55    
56    Bug fix for c-calls on x86 (having to do with how char- and
57    short-arguments are being handled).
58    
59    ----------------------------------------------------------------------
60    Name: Matthias Blume
61    Date: 2002/01/21
62    Tag: blume-20020121-ff
63    Description:
64    
65    Another day of fiddling with the FFI...
66    
67    1. Bug fix/workaround:  CKIT does not complain about negative array
68       dimensions, so ml-nlffigen has to guard itself against this possibility.
69       (Otherwise a negative dimension would send it into an infinite loop.)
70    
71    2. Some of the abstract types (light objects, light pointers, most "base"
72       types) in structure C are now eqtypes.
73    
74    3. Added constructors and test functions for NULL function pointers.
75    
76    ----------------------------------------------------------------------
77    Name: Matthias Blume
78    Date: 2002/01/18
79    Tag: blume-20020118-ready-for-new-release
80    Description:
81    
82    Made config/srcarchiveurl point to a new place.  (Will provide boot
83    files shortly.)
84    
85    Maybe we christen this to be 110.38?
86    
87    ----------------------------------------------------------------------
88    Name: Matthias Blume
89    Date: 2002/01/18
90    Tag: blume-20020118-more-ffifiddle
91    Description:
92    
93    Today's FFI fiddling:
94    
95      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
96        C.Set.* that use concrete (MLRep.*) arguments and results instead
97        of abstract ones.
98    
99      - Provided word-style bit operations etc. for "int" representation
100        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
101        and Long.
102    
103    ----------------------------------------------------------------------
104    Name: Matthias Blume
105    Date: 2002/01/18
106    Tag: blume-20020118-use-x86-fp
107    Description:
108    
109    Now that x86-fast-fp seems to be working, I turned it back on again
110    by default.  (Seems to work fine now, even with the FFI.)
111    
112    Other than that, I added some documentation about the FFI to
113    src/ml-nlffigen/README and updated the FFI test examples in
114    src/ml-nlffi-lib/Tests/*.
115    
116    ----------------------------------------------------------------------
117    Name: Allen Leung
118    Date: 2002/01/17
119    Tag: leunga-20020117-x86-fast-fp-call
120    Description:
121    
122       1. Fixed a problem with handling return fp values when x86's fast fp
123          mode is turned on.
124    
125       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
126          of %f32.
127    
128       3. Added a constructor INT32lit to the ast of MLRISC tools.
129    
130    ----------------------------------------------------------------------
131    Name: Matthias Blume
132    Date: 2002/01/16
133    Tag: blume-20020116-ffifiddle
134    Description:
135    
136    More fiddling with the FFI interface:
137    
138     - Make constness 'c instead of rw wherever possible.  This eliminates
139       the need for certain explicit coercions.  (However, due to ML's
140       value polymorphism, there will still be many cases where explicit
141       coercions are necessary.  Phantom types are not the whole answer
142       to modeling a subtyping relationship in ML.)
143    
144     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
145    
146     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
147       clumsy workaround for varargs problem.)
148    
149    ----------------------------------------------------------------------
150    Name: Lal George
151    Date: 2002/01/15
152    Tag: <none>
153    Description:
154    
155    1. Since COPY instructions are no longer native to the architecture,
156       a generic functor can be used to implement the expandCopies function.
157    
158    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
159       TEXT segment.
160    
161    ----------------------------------------------------------------------
162    Name: Matthias Blume
163    Date: 2002/01/15
164    Tag: blume-20020115-ffiupdates
165    Description:
166    
167    1. Fix for bug resulting in single-precision float values being returned
168       incorrectly from FFI calls.
169    
170    2. Small modifications to C FFI API:
171    
172        - memory-allocation routines return straight objects (no options)
173          and raise an exception in out-of-memory situations
174        - unsafe extensions to cast between function pointers and pointers
175          from/to ints
176        - added structure C_Debug as an alternative to structure C where
177          pointer-dereferencing (|*| and |*!) always check for null-pointers
178        - added open_lib' to DynLinkage;  open_lib' works like open_lib
179          but also takes a (possibly empty) list of existing library handles
180          that the current library depends on
181    
182    ----------------------------------------------------------------------
183    Name: Matthias Blume
184    Date: 2002/01/10
185    Tag: blume-20020110-newffigen
186    Description:
187    
188    1. Updates to portable graph code.
189    
190    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
191       more scalable now so that even huge interfaces such as the one
192       for GTK compile in finite time and space. :-)
193       See src/ml-nlffigen/README for details on what's new.
194    
195    ----------------------------------------------------------------------
196    Name: Lal George
197    Date: 2001/01/09 14:31:35 EST 2002
198    Tag: george-20011206-rm-native-copy
199    Description:
200    
201            Removed the native COPY and FCOPY instructions
202            from all the architectures and replaced it with the
203            explicit COPY instruction from the previous commit.
204    
205            It is now possible to simplify many of the optimizations
206            modules that manipulate copies. This has not been
207            done in this change.
208    
209    ----------------------------------------------------------------------
210    Name: Lal George
211    Date: 2001/12/06 16:50:13 EST 2001
212    Tag: george-20011206-mlrisc-instruction
213    Description:
214    
215    Changed the representation of instructions from being fully abstract
216    to being partially concrete. That is to say:
217    
218      from
219            type instruction
220    
221      to
222            type instr                              (* machine instruction *)
223    
224            datatype instruction =
225                LIVE of {regs: C.cellset, spilled: C.cellset}
226              | KILL of {regs: C.cellset, spilled: C.cellset}
227              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
228              | ANNOTATION of {i: instruction, a: Annotations.annotation}
229              | INSTR of instr
230    
231    This makes the handling of certain special instructions that appear on
232    all architectures easier and uniform.
233    
234    LIVE and KILL say that a list of registers are live or killed at the
235    program point where they appear. No spill code is generated when an
236    element of the 'regs' field is spilled, but the register is moved to
237    the 'spilled' (which is present, more for debugging than anything else).
238    
239    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
240    We used to generate:
241    
242            DEFFREG f1
243            f1 := f2 + f3
244            trapb
245    
246    but now generate:
247    
248            f1 := f2 + f3
249            trapb
250            LIVE {regs=[f1,f2,f3], spilled=[]}
251    
252    Furthermore, the DEFFREG (hack) required that all floating point instruction
253    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
254    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
255    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
256    this architecture).
257    
258    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
259    all the architectures. This will result in further simplification of the
260    register allocator that must be aware of them for coalescing purposes, and
261    will also simplify certain aspects of the machine description that provides
262    callbacks related to parallel copies.
263    
264    ANNOTATION should be obvious, and now INSTR represents the honest to God
265    machine instruction set!
266    
267    The <arch>/instructions/<arch>Instr.sml files define certain utility
268    functions for making porting easier -- essentially converting upper case
269    to lower case. All machine instructions (of type instr) are in upper case,
270    and the lower case form generates an MLRISC instruction. For example on
271    the alpha we have:
272    
273      datatype instr =
274         LDA of {r:cell, b:cell, d:operand}
275       | ...
276    
277      val lda : {r:cell, b:cell, d:operand} -> instruction
278        ...
279    
280    where lda is just (INSTR o LDA), etc.
281    
282    ----------------------------------------------------------------------
283    Name: Matthias Blume
284    Date: 2001/11/22 21:40:00 EST
285    Tag: Release_110_37
286    Description:
287    
288    Release 110.37.  This time for real.
289    
290    ----------------------------------------------------------------------
291    Name: Matthias Blume
292    Date: 2001/11/21 16:35:00 EST
293    Tag: blume-20011121-foot-in-mouth
294    Description:
295    
296    Removed the "Release_110_37" tag because of a serious bug.
297    This will be re-tagged once the bug is fixed.
298    
299    ----------------------------------------------------------------------
300    Name: Matthias Blume
301    Date: 2001/11/21 16:14:00 EST
302    Tag: blume-20011121-forgottenfile
303    Description:
304    
305    Forgot to add a file.  (Just a .tex-file -- part of
306    the CM manual source.)
307    
308    ----------------------------------------------------------------------
309    Name: Matthias Blume
310    Date: 2001/11/21 16:10:00 EST
311    Tag: blume-20011121-invalid_110_37
312    Description:
313    
314    Note: I removed the original tag "Release_110_37" from this commit
315          because we found a serious bug in all non-x86 backends.
316          - Matthias
317    
318    1. Modifications to the SML/NJ code generator and to the runtime system
319       so that code object name strings are directly inserted into code
320       objects at code generation time.  The only business the runtime system
321       has with this is now to read the name strings on occasions.
322       (The encoding of the name string has also changed somewhat.)
323    
324    2. CM now implements a simple "set calculus" for specifying export lists.
325       In particular, it is now possible to refer to the export lists of
326       other libraries/groups/sources and form unions as well as differences.
327       See the latest CM manual for details.
328    
329    3. An separate notion of "proxy" libraries has again be eliminated from
330       CM's model.  (Proxy libraries are now simply a special case of using
331       the export list calculus.)
332    
333    4. Some of the existing libraries now take advantage of the new set
334       calculus.
335       (Notice that not all libraries have been converted because some
336       of the existing .cm-files are supposed to be backward compatible
337       with 110.0.x.)
338    
339    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
340       "General.exnMessage"!  The former relies on a certain hook to be
341       initialized, and that often does not happen in the stand-alone case.)
342    
343    ----------------------------------------------------------------------
344    Name: Lal George
345    Date: 2001/11/21  13:56:18 EST
346    Tag: george-2001121-pseudo-ops
347    Description:
348    
349      Implemented a complete redesign of MLRISC pseudo-ops. Now there
350      ought to never be any question of incompatabilities with
351      pseudo-op syntax expected by host assemblers.
352    
353      For now, only modules supporting GAS syntax are implemented
354      but more should follow, such as MASM, and vendor assembler
355      syntax, e.g. IBM as, Sun as, etc.
356    
357    ----------------------------------------------------------------------
358    Name: Matthias Blume
359    Date: 2001/11/14 11:52:00 EST
360    Tag: blume-20011114-srcname
361    Description:
362    
363    1. Routed the name of the current source file to mlriscgen where it
364       should be directly emitted into the code object.  (This last part
365       is yet to be done.)
366    
367    2. Some cleanup of the pgraph code to make it match the proposal that
368       I put out the other day.  (The proposal notwithstanding, things are
369       still in flux here.)
370    
371    ----------------------------------------------------------------------
372    Name: Lal George
373    Date: 2001/11/14 09:44:04 EST
374    Tag:
375    Description:
376    
377      Fix for a backpatching bug reported by Allen.
378    
379      Because the boundary between short and long span-dependent
380      instructions is +/- 128, there are an astounding number of
381      span-dependent instructions whose size is over estimated.
382    
383      Allen came up with the idea of letting the size of span
384      dependent instructions be non-monotonic, for a maxIter
385      number of times, after which the size must be monotonically
386      increasing.
387    
388      This table shows the number of span-dependent instructions
389      whose size was over-estimated as a function of maxIter, for the
390      file Parse/parse/ml.grm.sml:
391    
392         maxIter            # of instructions:
393            10                      687
394            20                      438
395            30                      198
396            40                        0
397    
398      In compiling the compiler, there is no significant difference in
399      compilation speed between maxIter=10 and maxIter=40. Actually,
400      my measurements showed that maxIter=40 was a tad faster than
401      maxIter=10! Also 96% of the  files in the compiler reach a fix
402      point within 13 iterations, so fixing maxIter at 40, while high,
403      is okay.
404    
405    ----------------------------------------------------------------------
406    Name: Matthias Blume
407    Date: 2001/10/31 15:25:00 EST
408    Tag: blume-20011031-pgraph
409    Description:
410    
411    CKIT:
412    * Changed the "Function" constructor of type Ast.ctype to carry optional
413      argument identifiers.
414    * Changed the return type of TypeUtil.getFunction accordingly.
415    * Type equality ignores the argument names.
416    * TypeUtil.composite tries to preserve argument names but gives up quickly
417      if there is a mismatch.
418    
419    installation script:
420    * attempts to use "curl" if available (unless "wget" is available as well)
421    
422    CM:
423    * has an experimental implementation of "portable graphs" which I will
424      soon propose as an implementation-independent library format
425    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
426    
427    NLFFI-LIB:
428    * some cleanup (all cosmetic)
429    
430    NLFFIGEN:
431    * temporarily disabled the mechanism that suppresses ML output for
432      C definitions whose identifiers start with an underscore character
433    * generate val bindings for enum constants
434    * user can request that only one style (light or heavy) is being used;
435      default is to use both (command-line arguments: -heavy and -light)
436    * fixed bug in handling of function types involving incomplete pointers
437    * generate ML entry points that take record arguments (i.e., using
438      named arguments) for C functions that have a prototype with named
439      arguments
440      (see changes to CKIT)
441    
442    ----------------------------------------------------------------------
443    Name: Allen Leung
444    Date: 2001/10/27 20:34:00 EDT
445    Tag: leunga-20011027-x86-fast-fp-call
446    Description:
447    
448       Fixed the bug described in blume-20010920-slowfp.
449    
450       The fix involves
451          1. generating FCOPYs in FSTP in ia32-svid
452          2. marking a CALL with the appropriate annotation
453    
454    ----------------------------------------------------------------------
455    Name: Matthias Blume
456    Date: 2001/10/16 11:32:00 EDT
457    Tag: blume-20011016-netbsd
458    Description:
459    
460    Underscore patch from Chris Richards (fixing problem with compiling
461    runtime system under recent NetBSD).
462    
463    ----------------------------------------------------------------------
464    Name: Allen Leung
465    Date: 2001/10/12 17:18:32 EDT 2001
466    Tag: leung-20011012-x86-printflowgraph
467    Description:
468    
469    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
470    
471    ----------------------------------------------------------------------
472    Name: Lal George
473    Date: 2001/10/11 23:51:34 EDT
474    Tag: george-20011011-too-many-instrs
475    Description:
476    
477    The representation of a program point never expected to see more
478    than 65536 instructions in a basic block!
479    
480    ----------------------------------------------------------------------
481    Name: Lal George
482    Date: 2001/10/09 09:41:37 EDT
483    Tag: george-20011008-mlrisc-labels
484    Description:
485    
486    Changed the machine description files to support printing of
487    local and global labels in assembly code, based on host assembler
488    conventions.
489    
490    ----------------------------------------------------------------------
491    Name: Matthias Blume
492    Date: 2001/09/25 15:25:00 EDT
493    Tag: blume-20010925-exninfo
494    Description:
495    
496    I provided a non-hook implementation of exnName (at the toplevel) and
497    made the "dummy" implementation of exnMessage (at the toplevel) more
498    useful: if nothing gets "hooked in", then at least you are going to
499    see the exception name and a message indicating why you don't see more.
500    
501    [For the time being, programs that need exnMessage and want to use
502    ml-build should either use General.exnMessage (strongly recommended) or
503    refer to structure General at some other point so that CM sees a
504    static dependency.]
505    
506    [Similar remarks go for "print" and "use":  If you want to use their
507    functionality in stand-alone programs generated by ml-build, then use
508    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
509    
510    ----------------------------------------------------------------------
511    Name: Matthias Blume
512    Date: 2001/09/20 17:28:00 EDT
513    Tag: blume-20010920-slowfp
514    Description:
515    
516    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
517    
518    ----------------------------------------------------------------------
519    Name: Matthias Blume
520  Date: 2001/09/20 17:20:00 EDT  Date: 2001/09/20 17:20:00 EDT
521  Tag: blume-20010920-canonicalpaths  Tag: blume-20010920-canonicalpaths
522  Description:  Description:
# Line 33  Line 536 
536    
537  ----------------------------------------------------------------------  ----------------------------------------------------------------------
538  Name: Matthias Blume  Name: Matthias Blume
539    >>>>>>> 1.169
540  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
541  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
542  Description:  Description:

Legend:
Removed from v.937  
changed lines
  Added in v.1037

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0